A Bacterial Effector Targets Mad2l2, an APC Inhibitor, to Modulate Host Cell Cycling

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

A Bacterial Effector Targets Mad2L2, an APC Inhibitor, to Modulate Host Cell Cycling

Hiroki Iwai,1,7 Minsoo Kim,1,7 Yuko Yoshikawa,1 Hiroshi Ashida,1 Michinaga Ogawa,1 Yukihiro Fujita,1 Daniel Muller,1 Teruo Kirikae,4 Peter K. Jackson,5 Shuji Kotani,3,8 and Chihiro Sasakawa1,2,6,* 1Department of Microbiology and Immunology 2Department of Infectious Disease Control International Research Center for Infectious Diseases, Institute of Medical Science, University of Tokyo, 4-6-1, Shirokanedai, Minato-ku, Tokyo 108-8639, Japan 3Department of Molecular Cytogenetics, Medical Research Institute, Tokyo Medical and Dental University, 1-5-45, Yushima, Bunkyou-ku, Tokyo 113-0034, Japan 4Department of Infectious Diseases, Research Institute, International Medical Center of Japan, 1-21-1, Toyama, Shinjuku, Tokyo 162-8655, Japan 5Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA 6CREST, Japan Science and Technology Agency, Kawaguchi 332-0012, Japan 7These authors contributed equally to this work. 8Present address: Department of Internal Medicine, Arakawa Ryoyou Hospital, Arakawa, Tokyo, 116-0000, Japan *Correspondence: [email protected] DOI 10.1016/j.cell.2007.06.043

SUMMARY bloody and mucous-containing diarrhea. Shigella belong to the Escherichia coli family, but unlike other pathogenic The gut epithelium self-renews every several E. coli, Shigella have no adherence factors. Thus, follow- days, providing an important innate defense ing ingestion via the fecal-oral route, Shigella reach the co- system that limits bacterial colonization. Never- lon and rectum and translocate through the epithelial bar- theless, many bacterial pathogens, including rier via M cells overlying solitary lymphoid nodules. Once Shigella, efficiently colonize the intestinal epithe- they pass through the M cells, Shigella infect the resident lium. Here, we show that the Shigella effector macrophages and escape from the macrophages’ phago- somes into the cytoplasm, where they multiply and induce IpaB, when delivered into epithelial cells, causes rapid cell death (Sansonetti, 2004; Ogawa and Sasakawa, cell-cycle arrest by targeting Mad2L2, an ana- 2006). Shigella released from dead macrophages enter phase-promoting complex/cyclosome (APC) in- the surrounding enterocytes through the basolateral sur- hibitor. Cyclin B1 ubiquitination assays revealed face by inducing macropinocytosis. After a bacterium is that APC undergoes unscheduled activation surrounded by the membrane vacuole of an epithelial due to IpaB interaction with the APC inhibitor cell, it disrupts the vacuolar membrane and escapes into Mad2L2. Synchronized HeLa cells infected with the cytoplasm. Shigella multiply in the cytoplasm and Shigella failed to accumulate Cyclin B1, Cdc20, move intra- and intercellularly by inducing actin polymeri- and Plk1, causing cell-cycle arrest at the G2/M zation at one pole of the bacterium; this polymerization is phase in an IpaB/Mad2L2-dependent manner. required for cell-cell spreading (Cossart and Sansonetti, IpaB/Mad2L2-dependent cell-cycle arrest by 2004; Ogawa and Sasakawa, 2006). While establishing colonization, Shigella deliver a variety Shigella infection was also demonstrated in of virulence determinants, and the major determinants, rabbit intestinal crypt progenitors, and the IpaB- called effectors, are delivered through the type III secre- mediated arrest contributed to efficient coloni- tion system (TTSS) into host cells. For example, after Shi- zation of the host cells. These results strongly gella contact epithelial cells, the TTSS is stimulated and indicate that Shigella employ special tactics to delivers effectors including IpaA, IpaB, IpaC, IpgB1, influence epithelial renewal in order to promote IpgD, and VirA into the host cells; these effectors promote bacterial colonization of intestinal epithelium. bacterial invasion of host cells (Ogawa and Sasakawa, 2006; Parsot, 2005). One of these effectors, IpaB, plays multiple roles in promoting Shigella infection. The IpaB INTRODUCTION protein delivered into macrophages activates Caspase-1 and induces cell death (Hilbi et al., 1998, Chen et al., Shigella are highly adapted human pathogens that cause 1996; Zychlinsky et al., 1992). IpaB also interacts with bacillary dysentery, a disease that provokes severely CD44 on the epithelial cell membrane and stimulates

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 611 bacterial basolateral invasion (Lafont et al., 2002). Upon and it also targets mitotic Cyclin A and Cyclin B1, allowing contact of Shigella with the host plasma membrane, mitotic progression. Two proteins, Cdc20 and Cdc20 ho- IpaB, together with IpaC secreted at the tip of the TTSS molog 1 (Cdh1), bind directly to APC and thereby activate needle, acts as a membrane-pore-forming protein, medi- APC during mitosis and in the late mitotic and G1 phases, ating the translocation of effectors via the TTSS into the respectively (Peters, 2002). Mad2 and the spindle assem- host cell cytoplasm (Blocker et al., 1999). IpaB is also de- bly checkpoint pathway inhibit APCCdc20 activity in re- livered into the epithelial intracellular compartment during sponse to spindle damage (Nasmyth, 2005), and Early Shigella infection, but its biological role there is unknown. mitotic inhibitor 1 (Emi1) inhibits both Cdc20 and Cdh1 The intestinal epithelium self-renews every several in S and M phases (Hsu et al., 2002; Reimann et al., days, providing an important intrinsic defense system 2001a). However, additional APC regulators have been that limits bacterial colonization (Oswald et al., 2005; San- postulated to exist in order to ensure the destruction of sonetti, 2004). The fast turnover of intestinal epithelial cells APC substrates at the correct time and place. Indeed, re- forms crucial physical as well as functional barriers, and cent studies have identified p31-comet as a Mad2-binding this renewal is sustained by vigorous proliferation of epi- partner (Habu et al., 2002). In synchronized HeLa cells, thelial progenitors that migrate upwards from the bottom Mad2 forms a complex with Cdc20 during prometaphase of the intestinal crypts (Potten, 1998). Indeed, lesions in and metaphase; once the anaphase checkpoint is satis- the epithelial barrier are quickly repaired; if not, commen- fied, the Mad2-Cdc20 complex is dissociated by the bind- sal and pathogenic bacteria can enter the wound. Never- ing of Mad2 to p31-comet. Thus, p31-comet plays a criti- theless, many bacterial pathogens, including Shigella, are cal role in relieving inhibition of the Cdc20-containing APC capable of colonizing the intestinal barriers. As mentioned complex (Xia et al., 2004). Mad2L2, also referred to as above, Shigella (as well as many other mucosal Gram- Mad2B or hRev7 (Chen and Fang, 2001; Cheung et al., negative pathogenic bacteria) have developed a variety 2006; Pfleger et al., 2001) (Mad2L2 hereafter), which of tactics to manipulate host cell functions in order to shares 25% amino acid identity and 48% amino acid sim- avoid or overcome the innate host defense system. Re- ilarity to Mad2, was recently suggested to control APC ac- cent studies have indicated that a growing family of bac- tivation by Cdc20 (APCCdc20) and also to suppress APC terial toxins and effectors, termed ‘‘cyclomodulins,’’ inter- activation by Cdh1 (APCCdh1)(Chen and Fang, 2001; fere with the eukaryotic cell cycle, presumably to the Pfleger et al., 2001). Because Emi1 is capable of sup- pathogens’ benefit (Oswald et al., 2005). Some of these pressing APCCdh1 activation in human cell lines, it is as- cyclomodulins, such as the Cytolethal distending toxin sumed that Mad2L2’s function is similar to that of Emi1 (CDTs), are produced by a group of unrelated Gram- in controlling the APC activation required for the cell-cycle negative bacteria that includes S. dysenteriae (Nougayr- transition. ede et al., 2005). One of the CDTs produced by Campylo- In this study we investigated the route of Shigella infec- bacter jejuni possess a deoxyribonuclease I-like activity tion of the intestinal epithelium using rabbit ileal loops at and causes limited DNA damage when delivered into various stages of infection and noted that Shigella are ca- a host cell nucleus, eventually leading to chromatin dis- pable of accessing the intestinal crypt after inducing in- ruption (Lara-Tejero and Galan, 2000). Another cyclomo- flammatory responses in the intestinal tissue. We found dulin, Cycle inhibiting factor (Cif) inhibits host cell mitosis that the bacterial infection also reduces the proliferation after transfer from enteropathogenic E. coli via the TTSS of epithelial progenitors, as shown by Proliferating cell nu- (Marches et al., 2003). Cells transformed by Cif accumu- clear antigen (PCNA) staining. The effect was absent fol- late 4n DNA content and reinitiate DNA synthesis without lowing innoculation with a TTSS-deficient—and therefore dividing, resulting in 8–16n DNA content. G2 arrest of the noninvasive—Shigella mutant, such as the ipaB deletion host cell cycle is associated with the maintenance of mutant. More remarkably, we found that after transfer Cdk1, a cyclin-dependent kinase (Cdk), in its premitotic from Shigella into epithelial cells, intracellular IpaB can tyrosine-phosphorylated state (Marches et al., 2003). also be translocated into the host cell nucleus. This obser- However, the mechanisms underlying Cif-mediated vation led us to further investigate the role of IpaB in Shi- Cdk1 inactivation are unclear. Thus, some cell-cycle inhib- gella infection of intestinal epithelial cells. itors are anticipated to help prolong the pathogen’s presence by interfering with the rapid turnover of epithelial RESULTS cells or by harming the local epithelial barrier, allowing the entry of pathogenic bacteria into the subepithelial layer Shigella IpaB Targets Mad2L2 in Epithelial Cells (Lara-Tejero and Galan, 2002). at G2/M Phase Cell-cycle progression is stringently controlled by cell- To explore the role of IpaB delivered into epithelial cells cycle-specific proteolysis that involves the ubiquitination during Shigella infection of intestine, we performed yeast of target proteins by two main types of E3 ligase com- two-hybrid screening of HeLa cDNA library and found plexes, the anaphase-promoting complex (APC) and the that IpaB interacted with Mad2L2, an APC inhibitor Skp1-Cullin-F-box protein (SCF) complex (Vodermaier, (Figure 1A). Under these conditions, IpaB was incapable 2004). The APC is a multisubunit complex that targets of interacting with other APC inhibitors, such as Emi1 substrates for degradation only during mitosis and G1, and Mad2 (Figure 1A). The IpaB/Mad2L2 interaction was

612 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. Figure 1. IpaB Interacts with Mad2L2 (A) Binding of BD-IpaB1-312, BD-SipB1-319, AD-Mad2L2, AD-IpgC, AD-Emi1-N1-300, AD-Emi1-C250-447, and AD-Mad2 in a yeast two-hybrid assay. (B) IpaB bound to Mad2L2 in the GST-pull down assay (top). IpaB1-312 bound to MBP-Mad2L2 in the MBP pull-down assay but not to MBP-Emi1, MBP-Mad2, or MBP (bottom). Proteins that were pulled down were immunoblotted with antibodies speciﬁc for IpaB, IpaC, or IpaD.

(C) Cell lysates from HeLa cells infected with WT/IpaBFLAG were immunoprecipitated with anti-FLAG antibody or mouse IgG, and then immunoblotted with anti-Mad2L2 or -FLAG antibodies (left). Secretion of IpaBFLAG from Shigella was conﬁrmed using Congo red (CR)-treated supernatants from WT or WT/IpaBFLAG in immunoblots with anti-IpaB, -IpaC, or -FLAG antibodies (right). conﬁrmed by GST and MBP pull-down assays (Figure 1B). hours after infection of asynchoronously dividing HeLa To test the ability of IpaB to interact with endogenous cells, the length of time normally required for a HeLa cell

Mad2L2 we infected HeLa cells with WT/IpaBFLAG. This doubling, Mad2L2 was coimmunoprecipitated with Shigella strain expressed IpaBFLAG from the endogenous IpaBFLAG (Figure 1C, left). We next investigated the intra- 220 kb plasmid (Figure S1A) and carried DdapB so that in- cellular locations of IpaB and Mad2L2 in HeLa cells ex- tracellular Shigella would not kill the host cells in medium pressing GFP-IpaB1-312/Myc6-Mad2L2 or GFP-mock/ depleted of diaminopimelic acids (DAP). The expressed Myc6-Mad2L2 at the G1/S boundary induced by double IpaBFLAG conferred wild-type TTSS activity (Figure 1C, thymidine block (DTB) (Figure 2) Myc6-Mad2L2 and the right) and invasiveness (Figure S1B). In addition, the strain GFP control were detected in the nucleus and cytoplasm expressed AfaE (Aﬁmbrial adhesin E) from E. coli to boost from 0 to 15 hr. Although GFP-IpaB1-312 was detected in infection of HeLa cells (Philpott et al., 2000). Twenty-one the cytoplasm by 3 hr after DTB, IpaB and Mad2L2 were

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 613 Figure 2. IpaB Localizes in the Nuclei of HeLa Cells during G2/M Phase

GFP-IpaB1-312 and Myc6-Mad2L2 expressed in HeLa cells were visualized after G1/S release. HeLa-expressed GFP and Myc6-Mad2L2 were similarly examined. Cells were stained with TRITC-labeled anti-Myc antibody and Cy5-labeled anti-APC3 antibody. detected in the nucleus between 6 and 12 hr (Figure 2), When HeLa cell lysates were immunoprecipitated with suggesting that IpaB interacts with Mad2L2 in the G2/M anti-Mad2L2 or -Mad2 antibody, APC3, a component of phase. APC, was coprecipitated (Figure S3A, upper and bottom). Mad2, BubR1, Emi1, and RASSF1A are known APC in- When HeLa cell lysates were immunoprecipitated with hibitors in humans (Fang et al., 1998; Hsu et al., 2002; anti-APC3 antibody Emi1 was coprecipitated (Figure S3A, Song et al., 2004; Tang et al., 2001). Since the ability of middle). To identify the cell-cycle phase involved in the Mad2L2 to inhibit APC was previously determined only Mad2L2/APC interaction, HeLa cells synchronized at in Xenopus (Chen and Fang, 2001; Pﬂeger et al., 2001), the G1/S boundary were lysed at various times after we generated human anti-Mad2L2 antibody (Figure S2). DTB release and immunoprecipitated with anti-Mad2L2,

614 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. Figure 3. Shigella Cause Cell-Cycle Arrest (A) Rabbit ileal loops 6, 12, and 18 hr after inoculation with 2 3 108 GFP-Shigella were stained with rhodamine-phalloidin (red) and TO-PRO3 (blue). Bars, 100 mm. (B) The intestinal tissues were immunostained with FITC-labeled anti-PCNA antibody (green) and counterstained with rhodamine-phalloidin (red) and TO-PRO3 (blue). Bars, 100 mm. (C) Spherical HeLa cells after infection with WT, DipaB,oripaBWT (moi 80) (upper) and their percentages (bottom). *p < 0.0001. Data are represented as mean ± SEM. Cell-cycle stage distribution was determined by FACS (middle). (D) Spherical cells among synchronized HeLa cells infected with DipaB or WT (top). *p < 0.0001. Data are represented as mean ± SEM. HeLa spindle formation was impaired by WT infection but not by DipaB infection. DIC, a-tubulin (green) and DNA (blue). Bars, 10 mm. (E) Levels of cell-cycle-associated proteins in the cells used for (D) immunoblotting, with the speciﬁc antibodies indicated.

-APC3, or -Mad2 antibodies. The results showed that WT/pSU-GFP or the ipaB mutant (DipaB/pSU-GFP) was Mad2L2 interacted with APC primarily in G2/M phase inoculated into rabbit ileal loops, and intestinal tissue (Figure S3B). The cell-cycle distribution was monitored was examined 15 hr later. WT/pSU-GFP, but not DipaB/ by scoring the percentage of spherical cells (Figure S3C) pSU-GFP, destroyed the intestinal villi; on the other and using FACS analysis of DNA content (Figure S3D). hand, PCNA-positive cells in the crypts in Shigella- infected loops were barely detected (Figure 3B), implying Shigella Infection of Epithelial Cells Causes that Shigella interfer with the proliferation of epithelial Unscheduled Activation of APCCdh1 progenitors. HeLa cells were infected with or without To investigate whether Shigella affect the proliferation of dapB::Km/pIL14-Tp (bacteria expressing AfaE) (WT) in intestinal epithelial progenitors, YSH6000 (WT/pSU-GFP) the presence of DAP for 5 min. When HeLa cells cultured were visualized after inoculation into a rabbit ileal loop for 24 hr in DAP-free, gentamicin-containing MEM were in- (Figure 3A). Twelve and eighteen hours after inoculation, fected with WT and ipaBWT/dapB::Km/pIL14-Tp (ipaBWT) WT/pSU-GFP was detected at the bottom of crypts. (ipaB complement), approximately 20% of the infected

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 615 Figure 4. Shigella Cause Premature Activation of APCCdh1 (A) Analysis of Cyclin B1 ubiquitination activity in Shigella-infected HeLa cells. APC was im- munopuriﬁed from HeLa cells infected with WT and then incubated with biotinylated Cyclin B1 and ubiquitin (Left). Control mouse IgG or anti-APC3 immunoprecipitates from HeLa cells infected with WT were immunoblotted with anti-Cdh1 or -APC3 antibodies (right). (B) Effect of Shigella ipaB genotype on the half-lives of APC substrates after inoculation of asynchronous HeLa cells. Cychloheximide (CHX, 100 mg/ml) was added, and cell extracts were prepared at the time points indicated to investigate the half-lives of Cdc20, Plk1, Cyclin B1. ND indicates that changes in protein levels were not detected. (C) Effect of WT or DipaB infection of synchronous HeLa cells on the half-lives of the APC substrates.

cells, mostly those at G2/M phase, became spherical known substrates of APCCdh1, but not that of APCCdc20, (Figure 3C) and TUNEL-negative (data not shown), indicat- was low, and the level of Cdh1 decreased 6 hr after DTB ing that Shigella infection interfers with cell-cycle progres- release of the infected cells, suggesting that APCCdh1 un- sion. This was not the case with DipaB/dapB::Km/pIL14- dergoes unscheduled activation by Shigella (Figure 3E). Tp (DipaB)(Figure 3C). To characterize the spherical cells, This notion was also supported by WT Shigella (dapB+) in- DTB-synchronized cells at the G1/S boundary were infection of HeLa cells, where the accumulation of Cyclin B1 fected with WT or DipaB, and 5 min later the medium in WT infection was greatly decreased compared with that was replaced with DAP-free, gentamicin-containing of the noninfection control (Figure S4). MEM. WT infection resulted in more spherical cells than To demonstrate the activation of the APCCdh1 upon WT did DipaB infection (Figure 3D, upper panel). Spherical infection, we investigated the polyubiquitination level of cells with condensed chromatin emerged by 12 hr after Cyclin B1 (Figure 4A, left). Approximately 9 hr after DTB re- infection with the WT bacteria (Figure 3D, lower panel), lease, which corresponded to the G2/M phase, APCCdh1 whereas the cells generated by DipaB infection had the was activated, with the APC interacting with Cdh1 as ex- metaphase chromosome alignment and spindle organiza- amined by immunoprecipitation with anti-APC3 antibody tion (Figure 3D, lower panel). (Figure 4A, right). Indeed, when the level of APCCdh1 activ- Consistent with the results of Figure 2, spherical cells ity upon WT infection was examined by measuring the following WT infection were detected 6 hr after DTB re- half-lives of APC substrates such as Cdc20, Plk1, and Cy- lease (Figure 3D), suggesting that Shigella infection led clin B1, the half-lives were decreased relative to that of to M phase arrest. Immunoblotting of HeLa cells synchro- DipaB infection in either asynchronous (Figure 4B) or syn- nized in G1/S phase showed that although Cyclin B1, chronous HeLa cells (Figures 4C). Cdc20, and Plk1 accumulated approximately 9 hr after DTB release following DipaB infection, they were abol- IpaB and Mad2L2 Interaction Causes ished after WT infection (Figure 3E). WT infection was the Deregulation of APCCdh1 Activity also followed by an increase in phosphorylated (active) To deﬁne the IpaB portion involved in Mad2L2 binding, we APC, as well as a decrease in Cdh1 (an APC activator), performed a yeast two-hybrid assay with various IpaB compared with in the conditions following DipaB infection. truncations and found that Mad2L2 shared a binding site After WT infection, the accumulation of Cdc20 and Plk1, with IpgC (IpaB chaperone) on IpaB: residues 61–70

616 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. Figure 5. IpaB Suppresses Mad2L2 Inhi- bition of APCCdh1 (A) Mad2L2 bound Cdh1. 293T cells express-

ing GFP-Cdc20, GFP-Cdh1, Myc6-Mad2, or

Myc6-Mad2L2 and treated with Colcemid 32 hr after transfection were lysed 16 hr later and immunoprecipitated with anti-Myc antibody or rabbit IgG and immunoblotted with anti-GFP, -Mad2, or -Mad2L2 antibodies. (B) MBP-Mad2L2 bound to Cdc20 and Cdh1. MBP and MBP-Mad2L2 mixed with 35S- Cdc20 or 35S-Cdh1 were analyzed on a BAS2000. (C) MBP-Mad2L2 binding to 35S-Cdc20 or 35S- Cdh1 is inhibited by IpaB1-312WT in a dose- dependent manner. (D) In vitro ubiquitination assays for puriﬁed APC in the presence or absence of MBP- Mad2L2 or/and IpaB1-312WT. The APC was immunopuriﬁed from HeLa cell lysates and activated using recombinant Cdh1. The APC activated by Cdh1 was incubated with the N- terminal 100 residues of Cyclin B1-biotin. (E) The same assay as in (D) but performed with IpaB1-312N61A.

(Figure S5A). Because IpgC acts as the IpaB chaperone To confirm these findings, HeLa cells transfected with required for IpaB secretion via the TTSS (Page et al., Mad2L2-RNAi 96 hr prior to synchronization in G1/S 2001), we created Gal4 DBD-IpaB1-100 mutants that phase were re-transfected with Mad2L2-RNAi immedi- bound to AD-IpgC but not to AD-Mad2L2. An IpaB mutant ately after the first block, and the levels of Cdc20, Plk1, unable to bind MBP-Mad2L2, N61A (IpaB1-312N61A), was and Cyclin B1 were examined in cell lysates prepared be- further investigated (Figures S5B and S5C). By narrowing tween 0 and 15 hr after G1/S release (Figure S7). Although down the portions of Mad2L2 used in IpaB-binding exper- Luc-RNAi had no effect, Mad2L2 knockdown decreased iments, we found that residues 150–211 of Mad2L2 were the levels of Cdc20, Plk1, and Cyclin B1. The cellular levels involved in the interaction with IpaB (Figure S6). of pHistone H3 were also decreased by Mad2L2-RNAi ap- To examine the interaction between Mad2L2 and Cdh1, proximately 9 hr after DTB release (Figure S7A). In agree- we investigated 293T cells expressing GFP-Cdc20, GFP- ment with this, in Luc-RNAi-treated cells the peak of

Cdh1, Myc6-Mad2, or Myc6-Mad2L2 and found that more spherical cells occurred around 9 hr, opposed to a peak Cdh1 than Cdc20 was precipitated with Mad2L2 around 12 hr for the Mad2L2-RNAi-treated cells (Fig- (Figure 5A). When 35S-Cdc20 or 35S-Cdh1 was mixed ure S7B). FACS analysis also showed that 12 hr after with MBP-Mad2L2, Cdh1 and Cdc20 were precipitated DTB release the population at G2/M phase was increased by MBP-Mad2L2, but not by MBP alone (Figure 5B). in cells transfected with Mad2L2-RNAi relative to those When 35S-Cdc20 or 35S-Cdh1 were pulled down after in- treated with Luc-RNAi (Figure S7C). Indeed, when cubation with MBP-Mad2L2 in the presence of IpaB1- pHistone H3-positive cells transfected with Mad2L2- 312WT (at 1.5–24 mM), the amounts of 35S-Cdc20 and RNAi were examined by immunostaining approximately 35S-Cdh1 bound to MBP-Mad2L2 decreased as the 9 hr after DTB release, the percentage of positive cells amounts of IpaB1-312WT increased (Figure 5C), suggest- was greatly diminished compared with that of cells trans- ing that the IpaB/Mad2L2 interaction can activate fected with the Luc-RNAi control (Figure S7D). APCCdh1. Indeed, polyubiquitination of Cyclin B1 by Cdh1-activated APC was almost completely blocked by Involvement of the IpaB/Mad2L2 Interaction MBP-Mad2L2, but the blockage was absent in the pres- in Promoting Bacterial Colonization ence of IpaB1-312WT (Figure 5D). Abrogation of Cyclin of the Intestinal Epithelium B1 polyubiquitination by MBP-Mad2L2 was not prevented To establish the physiological relevance of IpaB/Mad2L2 by IpaB1-312N61A (Figure 5E), adding further evidence that interaction to Shigella-directed cell-cycle arrest, we the IpaB/Mad2L2 interaction promotes unscheduled replaced the ipaB gene of YSH6000 with ipaBN61A APCCdh1 activation. (ipaB::Km::ipaBN61A) and ipaBWT (ipaB::Km::ipaBWT) and

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 617 Figure 6. IpaB/Mad2L2-Dependent Cell- Cycle Arrest Is Exhibited during Shigella Infection (A) The competitive index (1.0, no attenuation) is a measure of the ability of ipaBN61A to colonize the rabbit ileal loops by comparing with WT. The competitive infections were carried out by the following combinations: WT/ipaBWT (n = 12), WT/ipaBN61A (n = 12), and WT/DipaB (n = 6). *p < 0.05. Data are represented as mean ± SEM. (B) Intestinal tissue immunostained with FITC- labeled anti-PCNA antibody (green) and counterstained with rhodamine-phalloidin (red) and TO-PRO3 (blue). Bars, 100 mm. (C) Time course of transient spherical pheno- type in synchronized HeLa cells after ipaBWT or ipaBN61A infection (moi 10). *p < 0.0001. Data are represented as mean ± SEM. (D) Lysates of HeLa cells stably expressing GFP, GFP-IpaB1-312WT, or GFP-IpaB1- 312N61A were immunoprecipitated with anti- GFP antibody and then immunoblotted with anti-GFP antibody. (E) Time course of spherical cell occurrence among synchronized HeLa cells stably expressing GFP, GFP-IpaB1-312WT, or GFP- IpaB1-312N61A. *p < 0.0001. **p < 0.05. Data are represented as mean ± SEM. (F) HeLa cells stably transfected with pEGFP, pEGFP-IpaB1-312WT, or pEGFP-IpaB1- 312N61A were released from thymidine block at the times indicated. The cell-cycle stages of the cells were determined by FACS.

confirmed that their TTSS activity and invasiveness were or ipaBN61A, the percentage of spherical cells at 12 hr after similar (Figures S8A, S8B, and S8C). We also confirmed release was lower with ipaBN61A than with ipaBWT (Fig- that the intracellular stabilities of IpaB and IpaBN61A in syn- ure 6C). Furthermore, when HeLa cells stably expressing chronized HeLa cells were similar (Figure S8D). We then GFP, GFP-IpaB1-312WT, or GFP-IpaB1-312N61A were prepared a series of bacterial mixtures composed of equal synchronized to the G1/S phase by DTB, after release the numbers of YSH6000 (WT)/ipaBWT, WT/ipaBN61A,or percentage of spherical cells was higher among cells ex- WT/DipaB, each of which were inoculated into rabbit intes- pressing GFP-IpaB1-312WT than those expressing GFP- tinal ileal loops. Eighteen hours after inoculation, we mea- IpaBN61A (Figure 6E). To confirm this, HeLa cells stably ex- sured bacterial numbers colonized per 0.5 cm2 of intestinal pressing GFP, GFP-IpaB1-312WT, or GFP-IpaB1-312N61A tissues. Although WT and ipaBWT colonized to a similar ex- were synchronized in the G1/S phase by DTB, and after tent, ipaBN61A showed less colonization (Figure 6A). More- release, the cells that had incorporated bromodeoxyuri- over, when 2 3 108 CFU of the strains were inoculated dine (BrdU) into their nuclei were identified by anti-BrdU- into different ileal loops, ipaBN61A and ipaBWT, but not FITC antibody (representing S phase cells) and propidium YSH6200 (the noninvasive mutant), caused severe inflam- iodide (PI) staining (representing G1 and G2/M phase cells). matory responses accompanying Shigella invasion of the Although the numbers of S- and G2/M-phased cells ex- intestinal tissues 12 hr after inoculation (data not shown). pressing GFP mock and GFP-IpaB1-312N61A were similar, Importantly, though PCNA-positive cells were rarely seen there were more such cells in cells expressing GFP-IpaB1- in the cryptic area after infection with ipaBWT, many were 312WT 12–15 hr after DTB release (Figure 6F). Thus, this se- observed after infection with ipaBN61A (Figure 6B), indicat- ries of experiments strongly suggested that the retardation ing that the IpaB/Mad2L2 interaction is important for pro- of rapid intestinal epithelial renewal, due to the interaction moting Shigella infection. When HeLa cells synchronized of IpaB/Mad2L2 during Shigella infection, is pivotal to pro- by DTB release were infected by Shigella carrying ipaBWT longing colonization within the intestinal epithelium.

618 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. DISCUSSION IpaB/Mad2L2 interaction participates in the activation of APCCdh1 at G2/M phase during Shigella infection. In the present study, we have provided convincing in vitro and in vivo evidence that Shigella can modulate the cell- Shigella Infection of Epithelial Cells Causes cycle progression of intestinal epithelial cells, and that Cell-Cycle Arrest the IpaB effector delivered via the TTSS into a host cell in- The intestinal epithelium self-renews every few days, and teracts with Mad2L2. This bacterial capability appears to its rapid turnover (including the tight sealing of epithelial affect epithelial cell turnover by slowing down cell renewal cell-cell junctions) is the most important intrinsic host de- during infection, enabling Shigella to persist longer in its fense against bacterial colonization. The ability of Shigella replicative niche. IpaB to inhibit cell-cycle progression thus seems to be a bacterial strategy to prolong local persistence by block- ing epithelial shedding. In addition, impairing the intestinal The IpaB/Mad2L2 Interaction Activates APCCdh1 epithelial integrity, which also occurs as a consequence of at G2/M Phase cell-cycle progression inhibition (Nougayrede et al., 2005; Mad2L2 was previously reported to be a Mad2-related Oswald et al., 2005), might further contribute to bacterial protein capable of binding APCCdh1 and inhibiting entry from the basolateral surface of epithelial cells, also APCCdh1 activation, though the precise role of Mad2L2 promoting bacterial colonization. Because at its initial in modulating cell-cycle progression remains to be eluci- stages Shigella infection of intestinal epithelium takes dated. Studies indicated that Mad2L2 plays a pivotal place via the M cells, and the bacterial ability to slow role in translesional DNA synthesis (TLS) in S phase cells cell-cycle progression was also anticipated to be less (Cheung et al., 2006; Murakumo et al., 2006), and that it effective on well-differentiated superficial epithelial cells also participates in the spindle assembly checkpoint, sim- (Sansonetti, 2004), we pursued whether Shigella could di- ilarly to Mad2 (Reimann et al., 2001b). In the present study, rectly access the intestinal cryptic area that is abundant in however, we did not observe GFP-IpaB1-312 within the epithelial progenitors. Using a rabbit ileal loop infection nucleus in S phase HeLa cells, implying that IpaB is not in- model (Figure 3A) and a guinea pig rectal infection model volved in TLS. Since IpaB lacks a nuclear localization sig- (Shim et al., 2007), we have shown that Shigella infection nal (NLS) and, at 62.2 kDa, is probably too large to enter can enter crypts and infect the cryptic epithelium by the nucleus via passive diffusion, we speculate that its en- 12 hr post-infection (Figure 3A). At that stage, though we try into the nucleus may occur by complex formation with detected few epithelial progenitors after WT Shigella Mad2L2, which does possess a putative NLS. Mad2L2 infection, epithelial progenitors remained abundant after shares significant amino acid identity (25%) and similarity infection with the ipaB mutant (Figure 3B). Likewise, the (48%) with Mad2 (Cahill et al., 1999). More recent studies ability of Shigella to induce cell-cycle arrest could be re- have indicated that Mad2L2 can suppress the activation produced using synchronized HeLa cells infected with of APCCdh1 (Pfleger et al., 2001). Emi1 is also known to WT but not ipaB mutant bacteria (Figure 3D). WT Shigella suppress APCCdh1 activation at G1/S phase, suggesting infection of HeLa cells generated more spherical cells with that Mad2L2 shows activity similar to Emi1 (Hsu et al., condensed chromatin than did the ipaB mutant 6 hr after 2002). APC activation by Cdc20 is predominant at M G1/S release; furthermore, the spherical cells were TUNEL phase, at which point the Mad2/Cdc20 interaction sup- negative, implying that the spherical cells were not a con- presses APC activation, while at M/G1 phase APC can sequence of apoptosis. Indeed, because the spherical be activated by Cdh1 (Peters, 2002). Mad2L2 is predom- HeLa cells caused by Shigella infection showed poor spin- inantly localized within the nuclei of BHK cells (Pfleger dle formation, we assumed that the cell-cycle arrest et al., 2001), and APC is also detected in the nuclei, sug- caused in the infected HeLa cells occurred prior to meta- gesting that the Mad2L2 is able to influence APC activity phase. Thus our data indicate that IpaB plays an important at M/G1 phase. The activity of APCCdh1 decreases at S role in causing cell-cycle arrest, although we have not ex- phase (Bashir et al., 2004; Wei et al., 2004), at which point cluded the possibility that other Shigella proteins might Emi1 participates in the inhibition of APC (Hsu et al., 2002), also play a role in affecting cell-cycle progression. raising the possibility that Mad2L2 also takes part in acti- vating APC at either S or G2/M phase. Therefore, using Shigella Infection of Epithelial Cells Preferentially cell lysates prepared from HeLa cells synchronized at Activates APCCdh1 G1/S phase, we investigated whether Mad2L2 could con- If the activation of APC in HeLa cells truly occurred as trol APC activity in M/G1, S, or G2/M phase. Our results of a consequence of the IpaB/Mad2L2 interaction during the immunoprecipitation assay with anti-Mad2L2 anti- Shigella infection, one would anticipate that the degrada- body clearly indicated that Mad2L2 does interact with tion of APCCdh1 substrates would also take place. We APC in HeLa cells at G2/M phase (Figure S3B). Consis- therefore investigated the levels of Cyclin B1, Cdc20, Cdh1 tently, when GFP-IpaB1-312 and Myc6-Mad2L2 were ec- and Plk1, known substrates for APC and, to a lesser topically expressed in synchronized HeLa cells, both sig- extent, for APCCdc20. Indeed, the levels of Cyclin B1, nals were detected within the nuclei at G2/M phase Cdc20, and Plk1 in HeLa cells 9 hr after G1/S release un- (Figure 2). Taken together, our results indicate that the derwent degradation after infecting with WT but not with

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 619 Figure 7. Proposed Model of the Role of IpaB in Cell-Cycle Progression Interfer- ence Mad2L2 interaction with APCCdh1 increases during G2/M (Figure S3B) and could therefore result in the accumulation of Cyclin B1, Cdc20, and Plk1 in G2/M (Figure S7A). Un- scheduled activation of APCCdh1 in G2/M by IpaB (Figure 3E) could explain the mitotic ab- normalities in Shigella-infected cells (Figure 3D) and, in turn, leads to premature degradation of Cyclin B1, Cdc20, and Plk1, a delay in mitotic progression, and faulty mitosis.

the ipaB mutant Shigella (Figure 3E). As Cdc20 and Plk1 mation inhibitor (Michel et al., 2001, 2004; De Antoni are substrates for APCCdh1 but not APCCdc20 (Castro et al., 2005; Luo et al., 2004). In contrast, knockdown of et al., 2005), the above results suggest that APCCdh1 acti- Mad2L2 results in a delay in cell-cycle progression (see vation, but not that of APCCdc20, occurred as the Shigella- Figure S7C). Although Mad2 also contributes to a delay induced unscheduled activation. In this regard, it is worth in cell-cycle progression, Mad2L2 can promote cell cy- noting that Shigella infection had a smaller effect on the cling as do APC inhibitors such as Emi1 and RASSF1A levels of Cyclin A than those of Cyclin B1, Cdc20, or (Hsu et al., 2002; Song et al., 2004). Importantly, upon Shi- Plk1 (Figure 3E). The fact that Cyclin A is an APCCdh1 sub- gella infection of HeLa cells, the phosphorylation of APC3 strate, and its rate of degradation is slower than that of was also detected, though the result was not dramatic other APCCdh1 substrates (Rape et al., 2006), may explain (Figure 3E). Although we do not understand how APC3 Shigella infection’s smaller effect on Cyclin A degradation phosphorylation is stimulated by Shigella infection, be- than on the other APCCdh1 substrates. In any case, our re- cause the levels of Cyclin B1 and Plk1 were greatly de- sults strongly indicate that an unscheduled activation of creased, the upregulation of phosphorylated APC3 does APCCdh1 takes place in HeLa cells at G2/M phase follow- not seem to be mediated by Cyclin B1-dependent or ing Shigella infection (Figure 4). Plk1 kinase; rather, other uncharacterized factor(s) may In addition to the activation of APC by Cdc20 and Cdh1, be involved in the phosphorylation of APC3. In HeLa cells the activation of APC is also accompanied by phosphory- infected with WT but not with ipaB mutant Shigella, the lation of APC3 (Kraft et al., 2003), an APC subunit, by other levels of Cdh1 were also remarkably decreased host factors. Knockout or -down of Mad2 causes early sis- (Figure 3E). As Cdh1 has been shown to undergo degra- ter chromatid segregation in the presence of spindle for- dation by APCCdh1 at G0 or G1 phase (Listovsky et al.,

620 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 2004), the decrease in Cdh1 after Shigella infection is also tion of intestinal epithelial renewal during Shigella infection consistent with bacterial infection-induced unscheduled prolongs bacterial colonization of the intestinal epithelium. activation of APCCdh1. To our knowledge, this is the first evidence that a bacterial TTSS-mediated effector, IpaB, targets epithelial cells and How Does Host Cell-Cycle Arrest by Shigella directly modulates their cell-cycle progression by target- Contribute to Bacterial Colonization? ing the activity of APC. The present study demonstrates that the cell-cycle progression modulated by APC’s ubiquitin ligase activity EXPERIMENTAL PROCEDURES can be slowed by the delivery of IpaB by Shigella into the host cells. How does this activity contribute to the bac- Yeast Two-Hybrid Assay The yeast two-hybrid screen was performed with the Matchmaker terial colonization of the intestinal epithelium? As noted GAL4 Two-Hybrid System 3 (Clontech), AH109 cotransformed with above, as the rapid turnover of epithelial layers limits bac- pGBKT7-ipaB1-936, a GAL4 DNA-binding domain vector, and terial colonization, it is possible that slowing this turnover a HeLa cDNA library cloned into a GAL4 activation domain vector, could prolong Shigella’s replicative foothold. Although the pACT2. situation is different than with Shigella, a similar hypothesis was recently postulated for Trichuris trichuria, a cecal- Pull-Down Assays GST-Mad2L2 bound to glutathione sepharose 4B was mixed with re- dwelling parasitic nematode. In the case of Trichuris,an combinant IpaB for 2 hr at 4C. After centrifugation, the beads were increase in the rate of epithelial cell turnover in murine washed three times with 1% Triton X-100-PBS and immunoblotted. large intestines acted as an ‘‘epithelial escalator’’ to effi- The MBP-Mad2L2 bound to the amylose resin was mixed with re- ciently expel Trichuris (Cliffe et al., 2005). To investigate combinant IpaB derivatives or 35S-labeled proteins and incubated this idea, we created IpaBN61A, a single amino acid- for 2 hr at 4C. Following centrifugation, the supernatant was dis- substituted IpaB mutant, which interacts more weakly carded and the beads were subjected to immunoblotting after washing with Mad2L2 (Figure S5) but still functions as a TTSS with 0.1% Tween 20-PBS eight times. translocator (Figures S8A and S8B) and supports bacterial Coimmunoprecipitations invasion of epithelial cells (Figure S8C). We compared this In our endogenous coimmunoprecipitation experiments, HeLa lysates mutant’s ability to deregulate APC activity via Mad2L2 to were spun at 15,000 rpm for 15 min at 4C. After adding 2 mgof that of WT IpaB in a Cyclin B1 ubiquitination assay. The primary antibody (anti-Mad2, -APC3, or -Mad2L2 antibodies)-protein results clearly showed that the inhibition of APCCdh1 by G-sepharose complex to the lysates, they were placed on ice for Mad2L2, as examined by the Cyclin B1 ubiquitination 2 hr. Mouse IgG was used as a negative control. Immunoprecipitates were washed three times with lysis buffer and then subjected to level, was reduced by the addition of IpaB but not immunoblotting. IpaBN61A into the assay medium (Figures 5D and 5E). Im- portantly, when a bacteria mixture composed of an equal Cell Synchronization number of Shigella expressing IpaB and Shigella express- HeLa G1/S cells were prepared by double thymidine block (DTB). The ing IpaBN61A was inoculated into rabbit ileal loops, the col- cells were first arrested with 2.0 mM thymidine for 20 hr, released into onization rate of Shigella expressing IpaB was much thymidine-free medium for 8 hr, and arrested again by incubation in 2.0 mM thymidine for 16 hr. Synchronous mitotic cells were obtained more efficient than that of Shigella expressing IpaBN61A by incubating for 16 hr in medium containing Colcemid (Demecolcine, (Figure 6A), further supporting the idea that the IpaB/ Sigma) at 150 ng/ml. Mad2L2 interaction is functionally important for bacterial colonization. In further support of our hypothesis, it has RNAi become increasingly evident that certain cyclomodulins, A small interfering RNA (CAGAAACGCAAGAAGUACA) duplex specific bacterial toxins, and effectors that modulate the eukary- for Mad2L2 repression was transfected into cells using Oligofect- otic cell cycle can target epithelial cells and control their AMINE (Invitrogen), according to the manufacturer’s instructions. cell-cycle progression (Nougayrede et al., 2005; Oswald Immunohistochemical Staining for PCNA et al., 2005). For example, Cytolethal-distending toxin To identify PCNA-positive crypt cells in the rabbit intestine, tissue (CDT), which is produced by a group of unrelated Gram- specimens fixed with 4% paraformaldehyde in PBS were frozen in liq- negative pathogenic bacteria, causes cell-cycle arrest in uid nitrogen and sectioned by a Leica cryostat (model CM 1900). The epithelial cells, though the targeted phase of the cell cycle sections were fixed for 4 min in acetone at 4C and then air-dried at varies between cell types (Lara-Tejero and Galan, 2000; room temperature. Sections were prepared and placed in 10 mM so- dium citrate buffer (pH 6.0), which was brought to a boil in a microwave Whitehouse et al., 1998). In these studies, the ability of oven, and the sections were allowed to cool to room temperature. The CDT to cause cell-cycle arrest raises the possibility that tissue was neutralized by two rinses in PBS-0.5% Tween 20 (PBS-T). prolonging local bacterial existence can be achieved by The sections were blocked for 30 min with PBS (pH 7.4) containing slowing epithelial renewal. 10% goat serum and incubated with primary antibody (anti-PCNA, In summary, our study has provided convincing evi- PC10) at a dilution of 1:100 in the above medium for 60 min. The sec- dence that Shigella influence epithelial cell-cycle progres- tions were then rinsed in PBS-T and incubated with secondary antibody (anti-mouse IgG-FITC, 1:100) for 60 min. sion by delivering IpaB into host cells. Once inside, IpaB interferes with the binding of Mad2L2 to Cdh1, leading Ubiquitination Assay to unscheduled activation of APC and subsequent Cyclin The ubiquitination assay was carried out as previously described (Ko- B1 degradation (Figure 7). We conclude that the retarda- tani et al., 1998). The N terminus of human Cyclin B1 cDNA, encoding

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 621 amino acid residues 1–100, was inserted into the BamHI-XhoI site of flexneri inserts IpaB and IpaC into host membranes. J. Cell Biol. 147, pET21a or NdeI site of pET42a and expressed in BL21 cells. The 683–693. His-tagged recombinant Cyclin B1 protein was purified and biotiny- Cahill, D.P., da Costa, L.T., Carson-Walter, E.B., Kinzler, K.W., Vogel- lated using a kit (Pierce). The purified APC was mixed in 30 mlof stein, B., and Lengauer, C. (1999). Characterization of MAD2B and 10 mM Tris (pH 7.6), 0.5 mM MgCl2, 1.5 mM DTT, 10 mM ATP, 5 mg other mitotic spindle checkpoint genes. Genomics 58, 181–187. recombinant E1, 1 mg/ml recombinant E2hC, 1 mg/ml recombinant E2H5B, 20 mg/ml ubiquitin (Sigma), 1 mg/ml recombinant Cdh1, and Castro, A., Bernis, C., Vigneron, S., Labbe, J.C., and Lorca, T. (2005). 10 mg/ml biotinylated Cyclin B1 and then incubated at 25C for 30 min. The anaphase-promoting complex: a key factor in the regulation of cell cycle. Oncogene 24, 314–325.

Competitive Colonization Assay Chen, J., and Fang, G. (2001). MAD2B is an inhibitor of the anaphase- Rabbit ileal loops were inoculated with 0.5 ml of a mixture (approxi- promoting complex. Genes Dev. 15, 1765–1770. mately 1:1) of WT (KmS) and ipaB mutant (KmR) Shigella (total inocu- Chen, Y., Smith, M.R., Thirumalai, K., and Zychlinsky, A. (1996). A bac- lum, 4 3 108 CFU in PBS), and each loop was treated twice with terial invasin induces macrophage apoptosis by binding directly to PBS containing 50 mg/ml gentamicin. Similar-sized tissue specimens ICE. EMBO J. 15, 3853–3860. obtained with an 8 mm diameter biopsy punch were washed with Cheung, H.W., Chun, A.C., Wang, Q., Deng, W., Hu, L., Guan, X.Y., PBS to eliminate residual gentamicin. The tissue samples were then Nicholls, J.M., Ling, M.T., Chuan Wong, Y., Tsao, S.W., et al. (2006). ground in ice-cold PBS, and a 1/10 dilution was prepared in tryptic Inactivation of human MAD2B in nasopharyngeal carcinoma cells soy broth (TCS) and incubated for 30 min at 37C with aeration. Serial leads to chemosensitization to DNA-damaging agents. Cancer Res. dilutions of this culture were then plated on agar and incubated over- 66, 4357–4367. night at 37C. CFUs were counted and the number of bacteria was calculated per 0.5 cm2 of intestinal tissue. Cliffe, L.J., Humphreys, N.E., Lane, T.E., Potten, C.S., Booth, C., and Grencis, R.K. (2005). Accelerated intestinal epithelial cell turnover: a new mechanism of parasite expulsion. Science 308, 1463–1465. Flow Cytometry Cell-cycle distribution was determined by propidium iodide (PI) or Cossart, P., and Sansonetti, P.J. (2004). Bacterial invasion: the para- PI/BrdU double staining. For PI/BrdU double staining, BrdU was digms of enteroinvasive pathogens. Science 304, 242–248. added to the culture medium 30 min prior to harvesting. BrdU incorpo- De Antoni, A., Pearson, C.G., Cimini, D., Canman, J.C., Sala, V., Nezi, ration was detected with a FITC-labeled mouse anti-BrdU antibody, L., Mapelli, M., Sironi, L., Faretta, M., Salmon, E.D., and Musacchio, A. and both DNA content and BrdU incorporation were analyzed by (2005). The Mad1/Mad2 complex as a template for Mad2 activation in a FACSCalibur flow cytometer. the spindle assembly checkpoint. Curr. Biol. 15, 214–225. Fang, G., Yu, H., and Kirschner, M.W. (1998). The checkpoint protein Statistical Analysis MAD2 and the mitotic regulator CDC20 form a ternary complex with Values are reported as means ± standard error of the mean. p values the anaphase-promoting complex to control anaphase initiation. were calculated by using a nonparametric one-tailed Mann-Whitney Genes Dev. 12, 1871–1883. U-test. Habu, T., Kim, S.H., Weinstein, J., and Matsumoto, T. (2002). Identifi- cation of a MAD2-binding protein, CMT2, and its role in mitosis. EMBO Supplemental Data J. 21, 6419–6428. Supplemental Data include Experimental Procedures, eight figures, Hilbi, H., Moss, J.E., Hersh, D., Chen, Y., Arondel, J., Banerjee, S., and Supplemental References and are available with this article online Flavell, R.A., Yuan, J., Sansonetti, P.J., and Zychlinsky, A. (1998). at http://www.cell.com/cgi/content/full/130/4/611/DC1/. Shigella-induced apoptosis is dependent on caspase-1 which binds to IpaB. J. Biol. Chem. 273, 32895–32900. ACKNOWLEDGMENTS Hsu, J.Y., Reimann, J.D., Sorensen, C.S., Lukas, J., and Jackson, P.K. (2002). E2F-dependent accumulation of hEmi1 regulates S phase entry We thank the members of the Sasakawa Laboratory, especially T. Su- by inhibiting APC(Cdh1). Nat. Cell Biol. 4, 358–366. zuki, S. Yoshida, and M. Fukumatsu, for their helpful advice. We are grateful to R.F. Whittier for critical reading of the manuscript. This Kotani, S., Tugendreich, S., Fujii, M., Jorgensen, P.M., Watanabe, N., work was supported by a Grant-in-Aid for the Scientific Research on Hoog, C., Hieter, P., and Todokoro, K. (1998). PKA and MPF-activated Priority Areas from the Ministry of Education, Culture, Sports, Science polo-like kinase regulate anaphase-promoting complex activity and and Technology of Japan (MEXT) and the Special Coordination Funds mitosis progression. Mol. Cell 1, 371–380. for Promoting Science from Japan Science and Technology Agency Kraft, C., Herzog, F., Gieffers, C., Mechtler, K., Hagting, A., Pines, J., (JSTA). M.K. is the recipient of the Japan Society for the Promotion and Peters, J.M. (2003). Mitotic regulation of the human anaphase- of Science (JSPS) postdoctoral fellowship for a foreign researcher. promoting complex by phosphorylation. EMBO J. 22, 6598–6609. P.K.J. is an employee of Genentech, Inc. Lafont, F., Tran Van Nhieu, G., Hanada, K., Sansonetti, P., and van der Goot, F.G. (2002). Initial steps of Shigella infection depend on the cho- Received: August 30, 2006 lesterol/sphingolipid raft-mediated CD44-IpaB interaction. EMBO J. Revised: March 29, 2007 21, 4449–4457. Accepted: June 21, 2007 Published: August 23, 2007 Lara-Tejero, M., and Galan, J.E. (2000). A bacterial toxin that controls cell cycle progression as a deoxyribonuclease I-like protein. Science 290, 354–357. REFERENCES Lara-Tejero, M., and Galan, J.E. (2002). Cytolethal distending toxin: Bashir, T., Dorrello, N.V., Amador, V., Guardavaccaro, D., and Pagano, limited damage as a strategy to modulate cellular functions. Trends Mi- M. (2004). Control of the SCF(Skp2-Cks1) ubiquitin ligase by the crobiol. 10, 147–152. APC/C(Cdh1) ubiquitin ligase. Nature 428, 190–193. Listovsky, T., Oren, Y.S., Yudkovsky, Y., Mahbubani, H.M., Weiss, Blocker, A., Gounon, P., Larquet, E., Niebuhr, K., Cabiaux, V., Parsot, A.M., Lebendiker, M., and Brandeis, M. (2004). Mammalian Cdh1/Fzr C., and Sansonetti, P. (1999). The tripartite type III secreton of Shigella mediates its own degradation. EMBO J. 23, 1619–1626.

622 Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. Luo, X., Tang, Z., Xia, G., Wassmann, K., Matsumoto, T., Rizo, J., and Potten, C.S. (1998). Stem cells in gastrointestinal epithelium: numbers, Yu, H. (2004). The Mad2 spindle checkpoint protein has two distinct characteristics and death. Philos. Trans. R. Soc. Lond. B Biol. Sci. 353, natively folded states. Nat. Struct. Mol. Biol. 11, 338–345. 821–830. Marches, O., Ledger, T.N., Boury, M., Ohara, M., Tu, X., Goffaux, F., Rape, M., Reddy, S.K., and Kirschner, M.W. (2006). The processivity of Mainil, J., Rosenshine, I., Sugai, M., De Rycke, J., and Oswald, E. multiubiquitination by the APC determines the order of substrate deg- (2003). Enteropathogenic and enterohaemorrhagic Escherichia coli radation. Cell 124, 89–103. deliver a novel effector called Cif, which blocks cell cycle G2/M transi- Reimann, J.D., Freed, E., Hsu, J.Y., Kramer, E.R., Peters, J.M., and tion. Mol. Microbiol. 50, 1553–1567. Jackson, P.K. (2001a). Emi1 is a mitotic regulator that interacts Michel, L., Diaz-Rodriguez, E., Narayan, G., Hernando, E., Murty, V.V., with Cdc20 and inhibits the anaphase promoting complex. Cell 105, and Benezra, R. (2004). Complete loss of the tumor suppressor MAD2 645–655. causes premature Cyclin B degradation and mitotic failure in human Reimann, J.D., Gardner, B.E., Margottin-Goguet, F., and Jackson, somatic cells. Proc. Natl. Acad. Sci. USA 101, 4459–4464. P.K. (2001b). Emi1 regulates the anaphase-promoting complex by Michel, L.S., Liberal, V., Chatterjee, A., Kirchwegger, R., Pasche, B., a different mechanism than Mad2 proteins. Genes Dev. 15, 3278– Gerald, W., Dobles, M., Sorger, P.K., Murty, V.V., and Benezra, R. 3285. (2001). MAD2 haplo-insufficiency causes premature anaphase and Sansonetti, P.J. (2004). War and peace at mucosal surfaces. Nat. Rev. chromosome instability in mammalian cells. Nature 409, 355–359. Immunol. 4, 953–964. Murakumo, Y., Mizutani, S., Yamaguchi, M., Ichihara, M., and Takaha- shi, M. (2006). Analyses of ultraviolet-induced focus formation of Shim, D.H., Suzuki, T., Chang, S.Y., Park, S.M., Sansonetti, P.J., Sasa- hREV1 protein. Genes Cells 11, 193–205. kawa, C., and Kweon, M.N. (2007). New animal model of shigellosis in the Guinea pig: its usefulness for protective efficacy studies. J. Immu- Nasmyth, K. (2005). How do so few control so many? Cell 120, 739– nol. 178, 2476–2482. 746. Song, M.S., Song, S.J., Ayad, N.G., Chang, J.S., Lee, J.H., Hong, H.K., Nougayrede, J.P., Taieb, F., De Rycke, J., and Oswald, E. (2005). Lee, H., Choi, N., Kim, J., Kim, H., et al. (2004). The tumour suppressor Cyclomodulins: bacterial effectors that modulate the eukaryotic cell RASSF1A regulates mitosis by inhibiting the APC-Cdc20 complex. cycle. Trends Microbiol. 13, 103–110. Nat. Cell Biol. 6, 129–137. Ogawa, M., and Sasakawa, C. (2006). Intracellular survival of Shigella. Cell. Microbiol. 8, 177–184. Tang, Z., Bharadwaj, R., Li, B., and Yu, H. (2001). Mad2-Independent inhibition of APCCdc20 by the mitotic checkpoint protein BubR1. Oswald, E., Nougayrede, J.P., Taieb, F., and Sugai, M. (2005). Bacte- Dev. Cell 1, 227–237. rial toxins that modulate host cell-cycle progression. Curr. Opin. Microbiol. 8, 83–91. Vodermaier, H.C. (2004). APC/C and SCF: controlling each other and the cell cycle. Curr. Biol. 14, R787–R796. Page, A.L., Fromont-Racine, M., Sansonetti, P., Legrain, P., and Parsot, C. (2001). Characterization of the interaction partners of Wei, W., Ayad, N.G., Wan, Y., Zhang, G.J., Kirschner, M.W., and Kae- secreted proteins and chaperones of Shigella flexneri. Mol. Microbiol. lin, W.G., Jr. (2004). Degradation of the SCF component Skp2 in cell- 42, 1133–1145. cycle phase G1 by the anaphase-promoting complex. Nature 428, Parsot, C. (2005). Shigella spp. and enteroinvasive Escherichia coli 194–198. pathogenicity factors. FEMS Microbiol. Lett. 252, 11–18. Whitehouse, C.A., Balbo, P.B., Pesci, E.C., Cottle, D.L., Mirabito, P.M., Peters, J.M. (2002). The anaphase-promoting complex: proteolysis in and Pickett, C.L. (1998). Campylobacter jejuni cytolethal distending mitosis and beyond. Mol. Cell 9, 931–943. toxin causes a G2-phase cell cycle block. Infect. Immun. 66, 1934– 1940. Pfleger, C.M., Salic, A., Lee, E., and Kirschner, M.W. (2001). Inhibition of Cdh1-APC by the MAD2-related protein MAD2L2: a novel mecha- Xia, G., Luo, X., Habu, T., Rizo, J., Matsumoto, T., and Yu, H. (2004). nism for regulating Cdh1. Genes Dev. 15, 1759–1764. Conformation-specific binding of p31(comet) antagonizes the function Philpott, D.J., Yamaoka, S., Israel, A., and Sansonetti, P.J. (2000). In- of Mad2 in the spindle checkpoint. EMBO J. 23, 3133–3143. vasive Shigella flexneri activates NF-kappa B through a lipopolysac- Zychlinsky, A., Prevost, M.C., and Sansonetti, P.J. (1992). Shigella charide-dependent innate intracellular response and leads to IL-8 flexneri induces apoptosis in infected macrophages. Nature 358, expression in epithelial cells. J. Immunol. 165, 903–914. 167–169.

Cell 130, 611–623, August 24, 2007 ª2007 Elsevier Inc. 623